Skip to main content

Table 3 Risk of on-road bicycle crashes in Auckland vs. the rest of New Zealand with stepwise adjustments

From: The role of multilevel factors in geographic differences in bicycle crash risk: a prospective cohort study

Models Additional variables in the model Beta estimates (SE) Hazard ratios (95% CI) % Attenuation (95% CI)a
1. Unadjusted   0.382 (0.073) 1.47 (1.27, 1.69)  
2. Model 1 + demographics Age 0.377 (0.073) 1.46 (1.26, 1.68)  
  Gender 0.375 (0.073) 1.46 (1.26, 1.68)  
  Ethnicity 0.370 (0.074) 1.45 (1.25, 1.67)  
  Education 0.369 (0.074) 1.45 (1.25, 1.67)  
  Body mass index 0.365 (0.074) 1.44 (1.25, 1.67) 4 (−1, 14)
3. Model 2 + residential factors NZDep 2006 scores 0.368 (0.074) 1.44 (1.25, 1.67)  
  Urban residence 0.292 (0.077) 1.34 (1.15, 1.56) 24 (11, 49)
4. Model 3 + cycling characteristics Years of cycling 0.291 (0.077) 1.34 (1.15, 1.56)  
  Time spent cycling 0.282 (0.077) 1.33 (1.14, 1.54)  
  % cycling off-road 0.245 (0.078) 1.28 (1.10, 1.49)  
  % cycling in the dark 0.217 (0.079) 1.24 (1.06, 1.45)  
  % cycling in a bunch 0.184 (0.079) 1.20 (1.03, 1.41)  
  Cycle to work 0.192 (0.079) 1.21 (1.04, 1.42)  
  Mainly use road bike 0.193 (0.080) 1.21 (1.04, 1.42)  
  Crash history 0.189 (0.080) 1.21 (1.03, 1.41) 51 (21, 74)
5. Model 4 + risk behaviours Use helmet 0.188 (0.080) 1.21 (1.03, 1.41)  
  Use fluorescent colours 0.188 (0.079) 1.21 (1.03, 1.41)  
  Use lights in the dark 0.189 (0.079) 1.21 (1.03, 1.41)  
  Use reflective materials in the dark 0.179 (0.080) 1.20 (1.02, 1.40)  
  Listen to music while riding 0.179 (0.080) 1.20 (1.02, 1.40) 53 (20, 72)
  1. a95% bootstrap confidence interval.